System Shock 3 - Why Warren Spector is making Games again

Gamasutra has interviewed Warren Spector about his return to the game industry:

The Comeback: Why Warren Spector is making games again

[…]

Why go back to full-time game development?

There are a bunch of reasons. You know, the first thing is, when I first started talking to the university, I told them I'd take a three-year commitment, because the game industry changes so quickly. I was worried that after three years, y'know, the relevance of what I know would start to diminish.

And I wanted to make sure I didn't become one of those teachers who used to make games, who used to know how games were developed and why. I knew I needed to keep my skills honed, so that was part of it. And part of it was just, y'know, I make games. It's kind of what I do. I've been getting the itch to make something. It's been coming on for a while now.

[…]

Let's dig into design, for a minute. How has the field of game design changed in the last few years, from your perspective, and where do you hope to take it?

I can't believe I'm about to say this -- I'll never work in this industry again -- but in the mainstream space I really haven't seen a whole lot of progress. It seems like we're getting more finely-tuned, prettier versions of games we've been playing for years.

Thank god for the indie space, there are people trying interesting things there. What I want to do, is I see a variety of places where we could make some strides that would help take games to the next level. The biggest one, for me, is more robust characters and character AI. We've gotten very good at combat AI, we've made great strides there, but I don't think we've done much in the world of non-combat AI and interacting with people -- human or otherwise. We haven't done a lot with conversation, and establishing emotional relationships with characters in games. So I'd very much like to play with that.

[…]

More information.

We haven't done a lot with conversation, and establishing emotional relationships with characters in games. So I'd very much like to play with that

How is that going to work with System Shock 3? Interactions with the Many or Shodan? Or will the crew not all be dead this time?

I think there's a ways we could go as well, in terms of empowering players to tell their own stories

More procedural open world stuff?

AI certainly hasn't developed much beyond say what Oblivion did. Not that this is surprising. AI is hard and just isn't there yet. But game developers do love to talk about it, making us all hopeful and dreamy, and then we can all be disappointed again after reading previews. Like with Bioshock Infinite or every Bethesda game.

Procedural quests are the horror.

AI is tough without a doubt. Conversational AI just seems out of alignment with SS1 and 2.

Perhaps. Although the ability to do more sophisticated verbal fencing and psychological back-and-forth with Shodan would be interesting. "Who are you calling insect, can-opener?"

Originally Posted by swaaye
AI certainly hasn't developed much beyond say what Oblivion did.

That's just scary.

Originally Posted by khaight
Perhaps. Although the ability to do more sophisticated verbal fencing and psychological back-and-forth with Shodan would be interesting. "Who are you calling insect, can-opener?"

*shrug* I don't know how much banter you expect there.

You are like the buzzing of flies to "him"…

Originally Posted by Cthulhufan
*shrug* I don't know how much banter you expect there.

You are like the buzzing of flies to "him"…

Perhaps our hero will be the nextgen Shodan portable?

Well, you can't question his experience or his contributions to several classic titles.

The one thing that has me a tiny bit concerned about Spector, is that as far as I recall, he took a rather rabid stance against video-game violence a few years ago. I can't remember the details (too lazy to look it up) but I think he rather strongly condemned violence in most forms of electronic entertainment.

I can't really see how that stance could fit well with a new System Shock game.

Don't forget System Shock was a result of a team of developers. Not only Warren. Lets hope, he doesn't mess up a perfect game.

I understood the spirit of what he was saying, but his comment about how the stakes are low was also a little disconcerting. Does he have the necessary hunger to make a great game? I also agree the conversational AI that interests him seems unrelated to past Shock games…

Originally Posted by Dave Allen
Perhaps our hero will be the nextgen Shodan portable?

The iShodan.

pibbur

https://www.youtube.com/watch?v=tvc7jzb9S8A

Warren Spector is right, the real reason to be gaming right now is because of the indie space.

I don't have to mention the last two games associated with his name.

I've seen too many has-beens and once-greats come back and disappoint to assume this is good news.

So far, I haven't been too impressed with Underworld Ascendant - though I still maintain hope it'll be pretty good.

However, I have this unfortunate feeling that people of a certain age and veterancy will inevitably struggle harder to maintain the passion and innovative inclination of youth.

This no matter how grand the intention.

Sort of agree with DArtagnan but there have been some good revivals lately with Pillars, WL2, etc. And then there are some big names in gaming that to me have always been overrated, like Cliffy and Carmack.

Maybe Spector's better talky talk can be projected onto speaking with the dead using some sort of Lovecraftian necromantic technology left behind by the old ones.

If he attempts a real run at solid AI development, I won't care what the setting is. I will throw money at the screen. You can iterate a lot on how to stuff meaningful choices into story-centric games, but the strength of the medium consistently rests on how mechanics are developed. It's almost a cliche now and a long ways from Extra Credits's pleas and even longer from Raph's Theory of Fun to understand games as their own art form in a way similar to Scott McCloud's "Understanding Comics". Throughout that time it's consistently been the indies that have pushed thoughtful mechanical design, usually because they have had to. We've gotten some really great iterations on big games too (open worlds and sandboxes really did use to be viewed with skepticism rather than treated as a money-making cliche). But the big money that has nothing to do with candy or birds has been continued iteration on graphical set pieces driving the horsepower available to consoles and PCs.

And I'm tired of it. It's old, tired, and stupid. I'm jaded and don't want ever smaller iterations on visual designs and filters when good art direction already does 80% of the work. We don't *need* pixel pushing anymore except for the rare game that's trying to kill the uncanny valley once and for all. And I just don't want to see any more asymptotic yet Quixotic tilts at that windmill. It's old.

I've since gotten myself into distributed high-speed computing via combinations of CUDA and MPI dedicated to solving scientific problems, usually thermal and aerospace simulations. It's ancillary, but related to AI development. I want to see all 4 cores on my computer utilized far more. I want to see every CUDA core on my graphics card working not just to push pixels but to try new things in game design through distributed horsepower that we haven't seen before. I want DICE and Bungie to swallow their martinis in shock at what can be done with a well-designed neural net or optimizations of a genetic algorithm. If anyone with a good sense of design can do it it's Phil + the old MIT crowd.

Regarding Oblivion: yeah radiant was a failure in AI. It's kind of ok, but only in an "mmo style quests are a nice digestif" way. I was tracking Bethsoft's claims and what happened at release and what I'm convinced happened is that they made a run at making it work and realized too late (after talking about it publicly) that the unpredictability of the simulations they wanted to do weren't going be fun. So they axed it and went with limited quest-style generation in Skyrim, going with their main strength in developing sheer volume. The lesson from this is not that AI is a natural failure, but that Bethsoft didn't dedicate space in their design document to make this a must-have priority. They gave it a try as a secondary objective. It didn't work so they cut and shipped the game; a good production decision from which we get bad conclusions about the possibilities of AI advances in mechanics.

It looks like Phil wants to give conversational AI development a significant run. It would work; we'll see. Maybe it won't be the wonderful utopia using every drop of computing power I can provide, but man I'd like to see some kind of success spark sudden realization from developers both big and small that there's possibilities in them thar hills.

It'd be a better use of engineering resources than decreasing returns on investment for graphics, that's for sure.

I just pray they don't plan to develop System Shock 3 with the Unity engine.

I mean, really.

The original System Shock all but revolutionised 3D engines - and it'd be painful to see them use Unity to make a modern version.

Unity is fine for many things - and I've even converted my own dungeon engine to it.

But System Shock should be about pushing the limits of immersion and sophisticated interaction.

I just don't see that happening in Unity without low performance or clunky gameplay.

I really hope they use UE4….. but I think they are smart enough to realize that themselves.